• Home
• About
• Writers
• Landscape
• Wizards Index
• Fraud
• Wanna Be Like Larry?

Enterprise Search: Is Keyword Search a Lycra-Spandex Technology?

I read a series of LinkedIn posts about why search may be an enterprise application flop. To access the meanderings of those who believe search is a young Bruce Jenner, you will have to sign up for LinkedIn and then wrangle an invitation to this discussion. Hey, good luck with this access to LinkedIn thing.

Over the years, enterprise search has bulked up. The keyword indexing has been wrapped in layers of helper code. For example, search now classifies, performs work flows operations, identifies entities, supports business intelligence dashboards, delivers self service Web help, handles Big Data, and dozens of other services.

Image Source: www.sochealth.co.uk.

I have several theories about this chubbification of keyword search. Let me highlight the thoughts that I jotted down as I worked through the “flop” postings on LinkedIn.

First, keyword search is not particularly useful to some people looking for information in an organization. The employee has to know what he or she needs and the terminology to use to unlock the secrets of the index. Add some time pressure and keyword search becomes infuriating. The fix, which began when Fulcrum Technologies pitched a platform approach to search, was to make search a smaller part of a more robust information management solution. You can still buy pieces of the original 1980s Fulcrum technology from OpenText today.

Second, system users continue to perceive results list as a type of homework. The employee has to browse the results list, click on documents that may contain the needed information, scan the document, identify the factoid or paragraph needed, copy it to another document, and then repeat the process. Employees want answers. What better way to deliver those answers than a “point and click” interface? Just pick what one needs and be done with the drudgery of the keyword search.

Third, professionals working in organizations want to find information from external sources like Web pages and blogs and from internal sources such as the server containing the proposals or president’s PowerPoint presentations. Enterprise search is presented as a solution to information access needs. The licensee quickly learns that most enterprise search systems require money, engineers, and time to set up so that content from disparate sources can be presented from a single interface. Again employees grouse when videos from YouTube and from the training department are not in the search results. Some documents containing needed information are not in the search system’s index but a draft version of the document is available via a Bing or Google search.

Fourth, the enterprise search system built on keywords lacks intelligence. For many vendors the solution is to add semantic intelligence, dynamic personalization which figures out what an employee needs by observing his information behaviors, and predictive analytics which just predicts what is needed for the company, a department and an individual.

Fifth, vendors have emphasized that a smart organization must have a taxonomy, a list of words and concepts tailored to the specific organization. These terms enrich the indexing of content. To make taxonomy management easy as pie, search vendors have tossed in editorial controls for indexing, classification, and hit boosting so that certain information appears whether the employee asked for the data or not.

In short order, the enterprise search system looks quite a bit like the “Obesity Is No Laughing Matter” poster.

This state of affairs is good for consulting engineers (SharePoint search, anyone?), mid tier consulting firm pundits, failed webmasters recast as search experts, and various hangers on. The obese enterprise search system is not particularly good for the licensing organization, the employees who are asked to use the system, or for the system administrators who have to shoehorn search into their already stuffed schedule for maintaining databases, accounting systems, enterprise resource planning, and network services.

Search is morbidly obese. No diet is going to work. The fix, based on the research conducted for my new monograph CyberOSINT is that a different approach is needed. Automated collection, analysis, and outputs are the future of information access.

Keyword search is a utility and available in NGIA systems. Unlike the obese keyword search systems, NGIA information access has been engineered to deliver more integrated services to users relying on mobile devices as well as traditional desktop computers.

Obese search is no laughing matter. One cannot make a utility into an NGIA system. However, and NGIA can incorporate search as a utility function. Keep this in mind if you are embracing Microsoft SharePoint-type systems. Net net: traditional enterprise search is splitting its seams, and it is unsightly.

Stephen E Arnold, March 3, 2015

Comments

Copyright © 2008 Beyond Search · ArnoldIT.com · Disclaimer · Revolution News theme by Brian Gardner · Log in